cover page labour discussion paper series - cream · labour discussion paper series cdp no 06/14...

23
Centre for Research and Analysis of Migration Department of Economics, University College London Drayton House, 30 Gordon Street, London WC1H 0AX L L a a b b o o u u r r Discussion Paper Series CDP No 06/14 From Sick Man of Europe to Economic Superstar: Germany’s Resurgent Economy Christian Dustmann, Bernd Fitzenberger, Uta Schönberg, Alexandra Spitz-Oener

Upload: vutuyen

Post on 30-Jun-2018

219 views

Category:

Documents


0 download

TRANSCRIPT

  • Centre for Research and Analysis of MigrationDepartment of Economics, University College LondonDrayton House, 30 Gordon Street, London WC1H 0AX

    LLLaaabbbooouuurrr

    Discussion Paper Series

    CDP No 06/14

    From Sick Man of Europe to

    Economic Superstar: Germanys

    Resurgent Economy

    Christian Dustmann, Bernd Fitzenberger,Uta Schnberg, Alexandra Spitz-Oener

  • Journal of Economic PerspectivesVolume 28, Number 1Winter 2014Pages 167188

    I n the late 1990s and into the early 2000s, Germany was often called the sick n the late 1990s and into the early 2000s, Germany was often called the sick man of Europe (for example, man of Europe (for example, Economist 2004), a phrase usually attributed 2004), a phrase usually attributed to comments by Czar Nicholas I of Russia about the troubles faced by the to comments by Czar Nicholas I of Russia about the troubles faced by the Ottoman Empire in the mid-19th century. Indeed, Germanys economic growth Ottoman Empire in the mid-19th century. Indeed, Germanys economic growth averaged only about 1.2percent per year from 1998 to 2005, including a recession averaged only about 1.2percent per year from 1998 to 2005, including a recession in 2003, and unemployment rates rose from 9.2percent in 1998 to 11.1percent in in 2003, and unemployment rates rose from 9.2percent in 1998 to 11.1percent in 2005 (according to World Bank data). Today, after the Great Recession, Germany is 2005 (according to World Bank data). Today, after the Great Recession, Germany is described as an economic superstar (for example, in the movie Made in Germany: described as an economic superstar (for example, in the movie Made in Germany: Europes Economic Superstar, http://fi lms.com/ItemDetails.aspx?TitleId=29218). Europes Economic Superstar, http://fi lms.com/ItemDetails.aspx?TitleId=29218). Germanys number of total unemployed fell from 5 million in 2005 to about Germanys number of total unemployed fell from 5 million in 2005 to about 3million in 2008, and its unemployment rate had declined to 7.7percent in 2010 3million in 2008, and its unemployment rate had declined to 7.7percent in 2010 (according to data from Germanys Federal Employment Agency, the Bundesagentur (according to data from Germanys Federal Employment Agency, the Bundesagentur fr Arbeit). In contrast to most of its European neighbors andthe United States, fr Arbeit). In contrast to most of its European neighbors andthe United States, Germany experienced almost no increase in unemployment during the Great Germany experienced almost no increase in unemployment during the Great Recession, despite a sharp decline in GDP in 2008 and 2009 (an episode discussed Recession, despite a sharp decline in GDP in 2008 and 2009 (an episode discussed

    From Sick Man of Europe to Economic Superstar: Germanys Resurgent Economy

    Christian Dustmann is Professor of Economics, University College London, London, Christian Dustmann is Professor of Economics, University College London, London, United Kingdom. Bernd Fitzenberger is Professor of Statistics and Econometrics, Univer-United Kingdom. Bernd Fitzenberger is Professor of Statistics and Econometrics, Univer-sity of Freiburg, Freiburg, Germany. Uta Schnberg is Associate Professor of Economics, sity of Freiburg, Freiburg, Germany. Uta Schnberg is Associate Professor of Economics, University College London, London, United Kingdom. Alexandra Spitz-Oener is Professor University College London, London, United Kingdom. Alexandra Spitz-Oener is Professor of Economics, Humboldt University Berlin, Berlin, Germany. Fitzenberger is also Research of Economics, Humboldt University Berlin, Berlin, Germany. Fitzenberger is also Research Associate at ZEW Mannheim, and Schnberg and Spitz-Oener are also Research Associ-Associate at ZEW Mannheim, and Schnberg and Spitz-Oener are also Research Associ-ates at the IAB, Nuremberg, Germany. Their email addresses are [email protected], ates at the IAB, Nuremberg, Germany. Their email addresses are [email protected], bernd.fi [email protected], [email protected], and alexandra.spitz-oenerbernd.fi [email protected], [email protected], and [email protected][email protected]. To access the Appendix, visithttp://dx.doi.org/10.1257/jep.28.1.167 doi=10.1257/jep.28.1.167

    Christian Dustmann, Bernd Fitzenberger, Uta Schnberg, and Alexandra Spitz-Oener

    http://films.com/ItemDetails.aspx?TitleId=29218mailto:[email protected]:[email protected]:[email protected]:[email protected]://dx.doi.org/10.1257/jep.28.1.167

  • 168 Journal of Economic Perspectives

    in Mller 2010; Burda and Hunt 2011). Germanys exports reached an all-time in Mller 2010; Burda and Hunt 2011). Germanys exports reached an all-time record of $1.738trillion in 2011, which is roughly equal to half of Germanys GDP, record of $1.738trillion in 2011, which is roughly equal to half of Germanys GDP, or 7.7percent of world exports. Even the euro crisis seems not to have been able to or 7.7percent of world exports. Even the euro crisis seems not to have been able to stop Germanys strengthening economy and employment.stop Germanys strengthening economy and employment.

    How did Germany, with the fourth-largest GDP in the world (after the United How did Germany, with the fourth-largest GDP in the world (after the United States, China, and Japan) transform itself from the sick man of Europe to an States, China, and Japan) transform itself from the sick man of Europe to an economic superstar in less than a decade? One common answer points to a series economic superstar in less than a decade? One common answer points to a series of legislative labor market reforms that started in the mid 2000s, the so-called Hartz of legislative labor market reforms that started in the mid 2000s, the so-called Hartz reforms. Another explanation focuses on the evolution of Germanys economy and reforms. Another explanation focuses on the evolution of Germanys economy and trade balance in the context of the eurozone. However, we will argue that these trade balance in the context of the eurozone. However, we will argue that these factors did not play a decisive role for the transformation of the German economy, factors did not play a decisive role for the transformation of the German economy, namely the restructuring of its labor market and the increase in competitiveness namely the restructuring of its labor market and the increase in competitiveness that has helped German exports. We instead present evidence that the specifi c that has helped German exports. We instead present evidence that the specifi c governance structure of the German labor market institutions allowed them to governance structure of the German labor market institutions allowed them to react fl exibly in a time of extraordinary economic circumstances, and that this react fl exibly in a time of extraordinary economic circumstances, and that this distinctive characteristic of its labor market institutions has been the main reason distinctive characteristic of its labor market institutions has been the main reason for Germanys economic success over the last decade.for Germanys economic success over the last decade.11

    We begin by arguing that the evolution of Germanys per unit labor coststhat We begin by arguing that the evolution of Germanys per unit labor coststhat is, labor costs relative to productivityin both the manufacturing sector and the is, labor costs relative to productivityin both the manufacturing sector and the other sectors in the economy has played an important role in the favorable evolution other sectors in the economy has played an important role in the favorable evolution of German tradable manufacturing industry. We then investigate the mechanisms of German tradable manufacturing industry. We then investigate the mechanisms that allowed for wage restraints and the dramatic decrease in real wages at the lower that allowed for wage restraints and the dramatic decrease in real wages at the lower end of the wage distribution.end of the wage distribution.

    The specifi c feature of the German system of industrial relations that we stress The specifi c feature of the German system of industrial relations that we stress is that it is not rooted in legislation, but instead is laid out in contracts and mutual is that it is not rooted in legislation, but instead is laid out in contracts and mutual agreements between the threemain actors in Germany: employer associations, trade agreements between the threemain actors in Germany: employer associations, trade unions, and works councils. The institutional setup of this system, which is dominated unions, and works councils. The institutional setup of this system, which is dominated by industry-wide wage bargaining, remained basically unchanged. However, many indi-by industry-wide wage bargaining, remained basically unchanged. However, many indi-cators demonstrate that it did change in the way it operates. For example, the share of cators demonstrate that it did change in the way it operates. For example, the share of German workers covered by any kind of union agreement has sharply declined, and German workers covered by any kind of union agreement has sharply declined, and the number of fi rm-level deviations from industry-wide union agreements has sharply the number of fi rm-level deviations from industry-wide union agreements has sharply increased since the mid 1990s. Overall, these gradual changes within the system led increased since the mid 1990s. Overall, these gradual changes within the system led to an unprecedented decentralization of the wage-setting process from the industry to an unprecedented decentralization of the wage-setting process from the industry level to the fi rm level. Alternatively, one may refer to this process as an increasing level to the fi rm level. Alternatively, one may refer to this process as an increasing localization of Germanys industrial relations.localization of Germanys industrial relations.

    The decentralization in wage setting in Germany is in contrast to many of its The decentralization in wage setting in Germany is in contrast to many of its neighbors where the statutory minimum wage is often high (relative to produc-neighbors where the statutory minimum wage is often high (relative to produc-tivity), where union wages and work hour regulations apply to all fi rms in the tivity), where union wages and work hour regulations apply to all fi rms in the

    1 Our argument is similar in spirit to that of Carlin and Soskice (2008, 2009), who argue that it is restruc-turing by Germanys private sector, using traditional German institutions based on employer-worker cooperation, and not government labor market and welfare state reforms that are to be credited for the German recovery.

  • Christian Dustmann, Bernd Fitzenberger, Uta Schnberg, and Alexandra Spitz-Oener 169

    industry, and where institutional change therefore requires broad consensus along industry, and where institutional change therefore requires broad consensus along the politicalspectrum.the politicalspectrum.

    We then turn to a discussion of why Germanys labor market experience has We then turn to a discussion of why Germanys labor market experience has been so distinctive within continental Europe. On the one side, the fall of the Berlin been so distinctive within continental Europe. On the one side, the fall of the Berlin Wall in 1989 and the dramatic cost of reunifi cation burdened the German economy Wall in 1989 and the dramatic cost of reunifi cation burdened the German economy in an unprecedented way, leading to a prolonged period of dismal macroeconomic in an unprecedented way, leading to a prolonged period of dismal macroeconomic performance. On the other side, it gave German employers access to neighboring performance. On the other side, it gave German employers access to neighboring East European countries that were formerly locked away behind the Iron Curtain, East European countries that were formerly locked away behind the Iron Curtain, and that were characterized by low labor cost, yet stable institutions and political and that were characterized by low labor cost, yet stable institutions and political structures. These factors changed the power equilibrium between employer and structures. These factors changed the power equilibrium between employer and employee associations and forced the latter to respond in a far more fl exible way employee associations and forced the latter to respond in a far more fl exible way than many would ever have expected. Finally, we discuss the relationship between than many would ever have expected. Finally, we discuss the relationship between our analysis of the fl exibility of Germanys labor market institutions and two other our analysis of the fl exibility of Germanys labor market institutions and two other events: Germanys Hartz reforms of 2003 and the arrival of the euro in 1999.events: Germanys Hartz reforms of 2003 and the arrival of the euro in 1999.

    How Did Germany Improve its Competitiveness?

    Relative Unit Labor CostIn Figure1, we plot the relative unit labor costs for a countrys overall economy In Figure1, we plot the relative unit labor costs for a countrys overall economy

    adjusted for the changing composition of the markets in which it competes, for a adjusted for the changing composition of the markets in which it competes, for a selection of countries, in dollar terms. This index is computed by the OECD based selection of countries, in dollar terms. This index is computed by the OECD based on year-to-year changes of unit labor costs and shows the relative change in the unit on year-to-year changes of unit labor costs and shows the relative change in the unit labor costs over time (normalized to 1995) translated into US dollars at the current labor costs over time (normalized to 1995) translated into US dollars at the current exchange rate compared to a weighted average of a countrys trading partners. The exchange rate compared to a weighted average of a countrys trading partners. The weights of the trading partners adjust annually to changes in trading patterns. An weights of the trading partners adjust annually to changes in trading patterns. An increase in this index indicates a deterioration of the competitive position. Adrop increase in this index indicates a deterioration of the competitive position. Adrop in this indexthat is, an improvement in competitivenessis caused by some in this indexthat is, an improvement in competitivenessis caused by some combination of threefactors: 1)a decrease in the wage per worker (or per hour); combination of threefactors: 1)a decrease in the wage per worker (or per hour); 2)an increase in productivity (per worker or per hour); and 3)a nominal deprecia-2)an increase in productivity (per worker or per hour); and 3)a nominal deprecia-tion of a countrys foreign exchange rate.tion of a countrys foreign exchange rate.

    Since 1995, Germanys competitive position has persistently improved, while Since 1995, Germanys competitive position has persistently improved, while the competitiveness of some of its main European trading partners has deteriorated the competitiveness of some of its main European trading partners has deteriorated (Spain and Italy) or remained close to the 1995 position (France). The competi-(Spain and Italy) or remained close to the 1995 position (France). The competi-tiveness of the United Kingdom has likewise deteriorated, although it improved tiveness of the United Kingdom has likewise deteriorated, although it improved dramatically between 2007 and 2009 due to the sharp depreciation of the British dramatically between 2007 and 2009 due to the sharp depreciation of the British pound against other currencies. The US economy also lost competitiveness rela-pound against other currencies. The US economy also lost competitiveness rela-tive to Germany in the late 1990s as the US dollar appreciated in value relative tive to Germany in the late 1990s as the US dollar appreciated in value relative to European currencies, but improved consistently after the 2001 recession, partly to European currencies, but improved consistently after the 2001 recession, partly achieved through a dollar depreciation (for instance, while the euro/dollar exchange achieved through a dollar depreciation (for instance, while the euro/dollar exchange rate was around 1 in 2001, it had depreciated to 0.8 in 2009). However, Germanys rate was around 1 in 2001, it had depreciated to 0.8 in 2009). However, Germanys gains in competitiveness with regard to France, Italy, and Spain cannot be due to gains in competitiveness with regard to France, Italy, and Spain cannot be due to currency depreciation (and in fact the euro appreciated relative to the currency of currency depreciation (and in fact the euro appreciated relative to the currency of most trading partners), because these countries all share the euro, and so it must most trading partners), because these countries all share the euro, and so it must

  • 170 Journal of Economic Perspectives

    have arisen because German wages grew at a slower pace than productivity relative have arisen because German wages grew at a slower pace than productivity relative to these other eurozone countries.to these other eurozone countries.

    Wage Trends and Wage InequalityFigure2 shows the evolution of real wages in West Germany since 1990. The Figure2 shows the evolution of real wages in West Germany since 1990. The

    fi gure illustrates the dramatic development in wage inequality in West Germany fi gure illustrates the dramatic development in wage inequality in West Germany over the past 15years or so (Dustmann, Ludsteck, and Schnberg 2009; see also over the past 15years or so (Dustmann, Ludsteck, and Schnberg 2009; see also Antonczyk, Fitzenberger, and Sommerfeld 2010; Card, Heining, and Kline 2013).Antonczyk, Fitzenberger, and Sommerfeld 2010; Card, Heining, and Kline 2013).22 Real wages at the 15th percentile fell dramatically from the mid 1990s onwards. Real wages at the 15th percentile fell dramatically from the mid 1990s onwards. From the early 2000s onwards, median real wages started to fall, and only wages From the early 2000s onwards, median real wages started to fall, and only wages at the top of the distribution continued to rise. Notice that all wage fi gures that at the top of the distribution continued to rise. Notice that all wage fi gures that we report stand for West Germany (although, henceforth, we refer to them as we report stand for West Germany (although, henceforth, we refer to them as Germany), because developments in East Germany are strongly affected by the Germany), because developments in East Germany are strongly affected by the transition after German unifi cation.transition after German unifi cation.

    2 Details on the wage data are in AppendixA available online with this journal at http://e-jep.org.

    Figure 1Evolution of Competition-Weighted Relative Unit Labor Costs, Selected Countries, 19942012

    Source: OECD Economic Indicators.Notes: This index accumulates the annual change in the relative unit labor costs of country i compared to a weighted average of its main trading partners where labor costs are translated into dollars and the weights are adjusted annually to the change in trade pattern. The annual change in logs is calculated as log( RULC it ) = log( ULC it e it ) ji

    g ij t1 log( ULC jt e jt ) where ULC it = ( w it L it )/ Y it is the unit labor

    cost for country i in period t, computed as the total wage bill w it L it divided by the value added of the countrys industry Y it . The unit labor costs are translated into US dollars using the exchange rate e it . Both the unit labor costs and the exchange rates are defi ned as index relative to some base year. The weighting scheme g ij t1 takes account of the structure of competition in both export and import markets of the goods sector of those countries, and it adjusts on a year-by-year basis. See OECD Economic Outlook (2012, Issue2, No.92) and OECD Economic Outlook Sources and Methods (http://www.oecd.org/eco/sources-and-methods) for details on the method of calculation.

    1.4

    1.3

    Inde

    x (1

    995

    = 1

    )

    1.2

    1.1

    1.0

    0.9

    0.8

    0.7

    United States

    United Kingdom

    Germany

    France

    Italy

    Spain

    19941995

    19961997

    19981999

    20002001

    20022003

    20042005

    20062007

    20082009

    20102011

    2012

    http://www.oecd.org/eco/sources-and-methodshttp://e-jep.org

  • From Sick Man of Europe to Economic Superstar: Germanys Resurgent Economy 171

    If the increase in wage inequality and the modest growth in wages overalland If the increase in wage inequality and the modest growth in wages overalland in particular the dramatic decline in real wages at the bottom of the wage distribu-in particular the dramatic decline in real wages at the bottom of the wage distribu-tionhas contributed to the favorable evolution of unit labor costs in Germany tionhas contributed to the favorable evolution of unit labor costs in Germany relative to the United States and other eurozone countries, then one should expect relative to the United States and other eurozone countries, then one should expect this development to have been particularly pronounced in the tradable manu-this development to have been particularly pronounced in the tradable manu-facturing sectorthe backbone of the German exporting industries accounting facturing sectorthe backbone of the German exporting industries accounting for 80percent of German exports. This insight turns out to hold true, but in an for 80percent of German exports. This insight turns out to hold true, but in an unexpectedway.unexpectedway.

    To further explore the increase in wage inequality, we classify sectors with To further explore the increase in wage inequality, we classify sectors with export volumes below the 25thpercentile of the distribution of export volumes in export volumes below the 25thpercentile of the distribution of export volumes in 1995 as nontradable sectors, and those with export volumes above this threshold 1995 as nontradable sectors, and those with export volumes above this threshold as tradable sectors. Tradable manufacturing are all those tradable sectors that as tradable sectors. Tradable manufacturing are all those tradable sectors that belong to the manufacturing sector, and tradable services are all other trad-belong to the manufacturing sector, and tradable services are all other trad-able sectors.able sectors.33 Figure 3 breaks down the evolution of real wages along the wage Figure 3 breaks down the evolution of real wages along the wage

    3 Details on the construction of these categories can be found in AppendixA, available with this paper at http://e-jep.org.

    Figure 2Indexed Wage Growth of the 15th, 50th, 85th Percentiles, West Germany, 19902008

    Notes: Calculations based on SIAB Sample for West German Full-Time Workers between 20 and 60years of age. The fi gure shows the indexed (log) real wage growth of the 15th, 50th, and 85thpercentiles of the wage distribution, with 1990 as the base year. Nominal wages are defl ated using the consumer price index (1995 = 100) provided by the German Federal Statistical Offi ce.

    .1

    .05

    0

    .05

    .1

    .15

    Inde

    xed

    wag

    e gr

    owth

    1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

    Year

    15th percentile

    50th percentile

    85th percentile

    http://e-jep.org

  • 172 Journal of Economic Perspectives

    distribution separately for the nontradable sector, tradable manufacturing sector distribution separately for the nontradable sector, tradable manufacturing sector (henceforth denoted as manufacturing), and tradable services sector. By this (henceforth denoted as manufacturing), and tradable services sector. By this measure, real wages in the manufacturing sector rose at all percentiles of the wage measure, real wages in the manufacturing sector rose at all percentiles of the wage distribution until the mid 2000s and afterwards continued to rise at the median distribution until the mid 2000s and afterwards continued to rise at the median and the 85th percentile. Germanys real wages in the nontradable sector hardly and the 85th percentile. Germanys real wages in the nontradable sector hardly increased at all at any part of the wage distribution during the 1990s and started to increased at all at any part of the wage distribution during the 1990s and started to decline from the early 2000s onwards even at the 85thpercentile, but particularly so decline from the early 2000s onwards even at the 85thpercentile, but particularly so at the 15thpercentile. The sharpest increase in inequality occurred in the tradable at the 15thpercentile. The sharpest increase in inequality occurred in the tradable service sector, where between 1990 and 2008 real wages did not show an increase at service sector, where between 1990 and 2008 real wages did not show an increase at the median, increased by 12percent at the 85thpercentile and declined by almost the median, increased by 12percent at the 85thpercentile and declined by almost

    Notes: Calculations based on SIAB Sample for West German Full-Time Workers between 20 and 60years of age. The fi gures show the indexed (log) real wage growth of the 15th, 50th, and 85thpercentiles of the wage distribution, with 1990 as the base year. Nominal wages are defl ated using the consumer price index (1995 = 100) provided by the German Federal Statistical Offi ce. PanelA shows the evolution of these fi gures for the nontradable sectors, panelB for tradable manufacturing, and panelC for tradable services. We classify sectors with export volumes below the 25thpercentile of the distribution of export volumes in 1995 as nontradable sectors, and those with export volumes above this threshold and that belong to the manufacturing sector as tradable manufacturing. The sectors above this threshold that do not belong to the manufacturing sector are classifi ed as tradable services.

    Figure 3Indexed Wage Growth of the 15th, 50th, 85th Percentiles, West Germany, by Sectors, 19902008

    .1

    .05

    0

    .05

    .1

    Inde

    xed

    wag

    e gr

    owth

    19901992

    19941996

    19982000

    20022004

    20062008

    Year

    15th percentile

    50th percentile

    85th percentile

    0

    .05

    .1

    .15

    .2

    Inde

    xed

    wag

    e gr

    owth

    19901992

    19941996

    19982000

    20022004

    20062008

    Year

    15th percentile

    50th percentile

    85th percentile

    .2

    .1

    0

    .1

    .2

    Inde

    xed

    wag

    e gr

    owth

    19901992

    19941996

    19982000

    20022004

    20062008

    Year

    15th percentile

    50th percentile

    85th percentile

    A: Nontradable Sectors B: Tradable Manufacturing

    C: Tradable Services

  • Christian Dustmann, Bernd Fitzenberger, Uta Schnberg, and Alexandra Spitz-Oener 173

    15percent at the 15thpercentile. At first glance, these figures do not seem to lend support to the hypothesis that wage restraint in the manufacturing sector was an important factor in improving competitiveness in that sector.

    Exports, Tradable Manufacturing, and Domestic InputsThe end product in manufacturing, however, contains a large share of inputs

    produced in other sectors: in Germany, the value added in manufacturing is only roughly one-third of the value of the end product, with the remainder of value added being contributed through inputs from other industries, either domesti-cally or from abroad (the literature so far has focused on Germanys imports of intermediate products from abroad, see Geishecker 2006; Sinn 2006; OECD 2007, chap.3; OECD 2012, chap.3). Hence, the manufacturing sector may have benefited from low wages in other domestic sectors and from cheap imports from abroad. In addition, Germanys manufacturing sector may have experienced increases in productivity which exceeded the increases in wages in the manufacturing sector.

    More detailed evidence suggests that both of these factors may be at play.4 In Germany, the manufacturing sector comprised 21.6percent of all jobs in 1995, but 17.7percent of all jobs in 2007, while the value added of this sector (in current prices) remained essentially unchanged at 22.8percent of all value added in 1995 compared with 22.7percent of value added in 2007. This pattern suggests larger productivity increases in the manufacturing sector than in the other sectors, where employment shares increased over the same period, with value added remaining roughly constant. This pattern is not uncommon across high-income countries.5 However, the share of manufacturing in output value (value of final products), as opposed to value added, rose steadily from 35percent of all output in 1995 to 39.3 percent of all output in 2007. This pattern reflects that the manufacturing sector indeed relies to an increasing extent on inputs from other domestic sectors and on imported inputs (because the share in final products has increased while the share in value added has remained the same), and may thus have benefited from the low wage growth in other domestic sectors and from cheaper imports.

    Digging down into the more detailed data, shown in Table1, the value of inputs over the value of output is nearly twice as high in manufacturing as in the other twosectors (66.1percent in 1995 versus 37.8percent in the tradable service sector) and this share increased by about 7 percentage points to 72.9 percent in 2007. The share of domestic inputs remained constant over the same period at about 51 percent. Thus, the increase in the share of inputs used by Germanys manu-facturing sector, relative to the output value in that sector, is driven by increased

    4 See Table1 and TableA1 in AppendixC available online with this paper at http://e-jep.org for details and data sources.5 Pilat, Cimper, Olsen, and Webb (2006) point out that the relatively fast productivity growth in manufac-turing is associated with relative declines of the prices for manufacturing products (this is Baumols cost disease). Thus, shares in value added at current prices understate the share of value added at constant prices in manufacturing to total value added at constant prices, which makes it remarkable that manu-facturing in Germany has retained its share in value added at current prices.

    http://e-jep.org

  • 174 Journal of Economic Perspectives

    use of inputs from abroad relative to inputs from domestic industries. However, use of inputs from abroad relative to inputs from domestic industries. However, even in 2007, 70percent of overall inputs in Germanys manufacturing sector were even in 2007, 70percent of overall inputs in Germanys manufacturing sector were domestically produced. Thus, the argument that Germanys manufacturing sector domestically produced. Thus, the argument that Germanys manufacturing sector has become nothing more than an assembly place for foreign produced inputs (for has become nothing more than an assembly place for foreign produced inputs (for example, Sinn 2006) is unjustifi ed. In fact, while German manufacturing has made example, Sinn 2006) is unjustifi ed. In fact, while German manufacturing has made increasing use of imported inputs, the share of domestic inputs in manufacturing increasing use of imported inputs, the share of domestic inputs in manufacturing fi nal output value had remained high and relatively stable between 1995 and 2007.fi nal output value had remained high and relatively stable between 1995 and 2007.

    To what extent have Germanys domestic inputs contributed to competitiveness To what extent have Germanys domestic inputs contributed to competitiveness in its export-oriented manufacturing sector and the two other sectors? In Figure4, in its export-oriented manufacturing sector and the two other sectors? In Figure4, we plot the evolution of unit labor costs in the threesectors, where industries are we plot the evolution of unit labor costs in the threesectors, where industries are weighted with respect to their exports for the twotradable sectors.weighted with respect to their exports for the twotradable sectors.66 When computing When computing unit labor costs, we fi rst consider only the value added in the sector, as denoted by unit labor costs, we fi rst consider only the value added in the sector, as denoted by solid lines in Figure4. We then consider fi nal output value in the sector, which is the solid lines in Figure4. We then consider fi nal output value in the sector, which is the sum of value added in the sector and all inputs into the sector denoted by dotted lines sum of value added in the sector and all inputs into the sector denoted by dotted lines in Figure 4. This index (Unit Labor Costs: End Products) incorporates gains in in Figure 4. This index (Unit Labor Costs: End Products) incorporates gains in competitiveness in a sector due to the usage of inputs from other domestic sectors. We competitiveness in a sector due to the usage of inputs from other domestic sectors. We also plot median real wages, adjusted using Germanys Consumer Price Index, for the also plot median real wages, adjusted using Germanys Consumer Price Index, for the threesectors. While real wage growth in the manufacturing sector is relatively modest, threesectors. While real wage growth in the manufacturing sector is relatively modest,

    6 Details on how unit labor costs are calculated can be found in AppendixA and in AppendixC, which are available online with this article at http://e-jep.org.

    Table 1Evolution of the Share of Value of Total Inputs and Domestic Inputs over the Value of Output, Overall and by Sector, 19952007

    Overall Nontradable sectors Tradable manufacturing Tradable services

    Panel A: Value of Total Inputs/Output Value1995 48.2% 39.9% 66.1% 37.9%2000 51.0% 37.9% 70.1% 41.4%2007 53.2% 38.2% 72.9% 41.6%

    Panel B: Value of Domestic Inputs/Output Value1995 39.8% 35.3% 51.7% 32.4%2000 40.3% 32.2% 51.7% 34.8%2007 40.5% 32.1% 51.2% 34.2%

    Panel C: Value of Domestic Inputs/Value of Total Inputs1995 82.6% 88.3% 78.1% 85.6%2000 79.0% 84.9% 73.7% 84.0%2007 76.1% 83.9% 70.3% 82.2%

    Notes: Calculations based on input-output statistics from the German Statistical Offi ce (Fachserie 18, Reihe2, Years:1995 2007). We classify sectors with export volumes below the 25thpercentile of the distribution of export volumes in 1995 as nontradable sectors and those with export volumes above this threshold and that belong to the manufacturing sector as tradable manufacturing. The sectors above this threshold that do not belong to the manufacturing sector are classifi ed as tradable services.

    http://e-jep.org

  • From Sick Man of Europe to Economic Superstar: Germanys Resurgent Economy 175

    at about 8.2percent over the 11-year period, in the other twosectors average wages at about 8.2percent over the 11-year period, in the other twosectors average wages fell in real terms by 1.2 and 4.1percent, respectively, over this time period.fell in real terms by 1.2 and 4.1percent, respectively, over this time period.

    As visible in the fi gure, domestic unit labor cost for total production in manu-As visible in the fi gure, domestic unit labor cost for total production in manu-facturing, taking account of inputs produced in other sectors (end products), facturing, taking account of inputs produced in other sectors (end products), declined far more rapidly than unit labor costs in value addeda decline that cannot declined far more rapidly than unit labor costs in value addeda decline that cannot be explained by the increase in the share of imported inputs in total output value. be explained by the increase in the share of imported inputs in total output value. Moreover, unit labor costs in end products start to decline at the start of the observa-Moreover, unit labor costs in end products start to decline at the start of the observa-tion period in 1995, while unit labor costs in value added decrease rapidly only from tion period in 1995, while unit labor costs in value added decrease rapidly only from 2003 onwards when mean wages, and in particular wages at the 15thpercentile of 2003 onwards when mean wages, and in particular wages at the 15thpercentile of the wage distribution start to decrease in real terms (as shown earlier in Figure3).the wage distribution start to decrease in real terms (as shown earlier in Figure3).

    Thus, Germanys manufacturing sector improved competitiveness in several Thus, Germanys manufacturing sector improved competitiveness in several ways. First, manufacturing drew on inputs from domestically provided nontradable ways. First, manufacturing drew on inputs from domestically provided nontradable and especially tradable services, where real wages fell between 1995 and 2007. Second, and especially tradable services, where real wages fell between 1995 and 2007. Second, the decline in unit labor costs, coupled with the increase in mean real wages in the decline in unit labor costs, coupled with the increase in mean real wages in manufacturing, implies that productivity increases in the manufacturing sector have manufacturing, implies that productivity increases in the manufacturing sector have

    Notes: The fi gures show indexed real mean daily wages by sector (base year 1995 = 100). Nominal wages are defl ated using the consumer price index (1995 = 100) provided by the German Federal Statistical Offi ce. The data underlying these indices are in columns(1), (4), and (7) of TableA2. The fi gure also shows indexed unit labor costs both Value added and End product by sector. The data underlying these indices are in columns(2), (3), (5), (6), (8), and (9) of TableA2 of the online Appendix. TableA2 includes a detailed description of data and methods of calculations.

    Figure 4Evolution of Real Daily Wages and Unit Labor Costs by Sector, 19952007

    110105100959085807570 1995

    19971999

    20012003

    20052007

    Real daily wages

    Unit labor costs: Value added

    Unit labor costs: End product

    110105100959085807570 1995

    19971999

    20012003

    20052007

    Real daily wages

    Unit labor costs: Value added

    Unit labor costs: End product

    110105100959085807570 1995

    19971999

    20012003

    20052007

    Unit labor costs: End product

    Unit labor costs: Value added Real daily wages

    Inde

    xede

    d w

    age

    grow

    th

    Inde

    xede

    d w

    age

    grow

    th

    A: Nontradable Sectors B: Tradable Manufacturing

    C: Tradable Services

  • 176 Journal of Economic Perspectives

    outpaced wage increases in that sector. In comparison, total unit labor costs fell less in outpaced wage increases in that sector. In comparison, total unit labor costs fell less in the nontradable sectors (minus 22.2percent) and much less in the tradable services the nontradable sectors (minus 22.2percent) and much less in the tradable services (minus 9.7percent), even though nominal wages grew much less in these twosectors (minus 9.7percent), even though nominal wages grew much less in these twosectors compared to tradable manufacturing. Note also that productivity increases in the compared to tradable manufacturing. Note also that productivity increases in the manufacturing sector have exceeded the increases in the two other sectors. Finally, to manufacturing sector have exceeded the increases in the two other sectors. Finally, to increase the competitiveness of its own fi nal products, the manufacturing sector has increase the competitiveness of its own fi nal products, the manufacturing sector has made increased use of trade integration with Eastern European countries through made increased use of trade integration with Eastern European countries through inputs imported from abroad, and far more so than other European countries. inputs imported from abroad, and far more so than other European countries. Theseinputs made up 14.5percent of total output in the manufacturing sector in 1995 Theseinputs made up 14.5percent of total output in the manufacturing sector in 1995 and 21.5percent in 2007. Calculating the outsourcing indicator suggested by Egger and and 21.5percent in 2007. Calculating the outsourcing indicator suggested by Egger and Egger (2003, p. 642) for Germany, France, and Italy regarding imported inputs Egger (2003, p. 642) for Germany, France, and Italy regarding imported inputs from Poland, Hungary, and the Czech and the Slovak Republics, using data from from Poland, Hungary, and the Czech and the Slovak Republics, using data from the OECD Input-Output-Tables (at http://www.oecd.org/trade/input-outputtablesthe OECD Input-Output-Tables (at http://www.oecd.org/trade/input-outputtables.htm) and OECD International Trade and Balance of Payments Statistics (at http://.htm) and OECD International Trade and Balance of Payments Statistics (at http://www.oecd.org/std/its/), shows that in the year 2000, imported inputs from these www.oecd.org/std/its/), shows that in the year 2000, imported inputs from these fourcountries amounted to about 8.5percent of inputs in Germany, compared to fourcountries amounted to about 8.5percent of inputs in Germany, compared to 2.5percent in Italy and 1.9percent in France (relative toGDP).2.5percent in Italy and 1.9percent in France (relative toGDP).

    The Increase in Competitiveness and Germanys Labor Market Institutions

    The movements in German wages, within and across sectors, belie the common The movements in German wages, within and across sectors, belie the common belief that Germanys labor market institutions are overly rigid. Instead, we argue that belief that Germanys labor market institutions are overly rigid. Instead, we argue that the specifi c governance structure of the German system of industrial relations offers the specifi c governance structure of the German system of industrial relations offers various margins of fl exibility. In the early to mid 1990s, these institutions allowed for various margins of fl exibility. In the early to mid 1990s, these institutions allowed for an unprecedented increase in the decentralization (localization) of the process that an unprecedented increase in the decentralization (localization) of the process that sets wages, hours, and other aspects of working conditions, from the industry- and sets wages, hours, and other aspects of working conditions, from the industry- and region-wide level to the level of the single fi rm or even the single worker, which in region-wide level to the level of the single fi rm or even the single worker, which in particular helped to bring down wages at the lower end of the wage distribution. particular helped to bring down wages at the lower end of the wage distribution. This decentralization took place even though the institutional setup of the domi-This decentralization took place even though the institutional setup of the domi-nating system of industry-wide wage bargaining basically remainedunchanged.nating system of industry-wide wage bargaining basically remainedunchanged.

    The specifi c feature which we stress here is that the governance structure of The specifi c feature which we stress here is that the governance structure of the German system of industrial relations is not rooted in legislation and is not the German system of industrial relations is not rooted in legislation and is not governed by the political process, but instead is laid out in contracts and mutual governed by the political process, but instead is laid out in contracts and mutual agreements between the three main labor market parties: trade unions, employer agreements between the three main labor market parties: trade unions, employer associations, and works councils (the worker representatives who are typically associations, and works councils (the worker representatives who are typically present in medium-sized and large fi rms).present in medium-sized and large fi rms).77 For this reason, Germany was in the For this reason, Germany was in the position to react in an unprecedented way to the challenges of the early 1990s.position to react in an unprecedented way to the challenges of the early 1990s.

    7 Works councils have to be set up in establishments with more than fi veemployees when demanded so by the employees. About 92percent of employees in establishments that have more than 50employees work in establishments with a works council, but only 18percent of employees in establishments that are smaller (Addison. Schnabel, and Wagner 1997; Beckmann, Fhr, and Krkel 2010).

    http://www.oecd.org/trade/input-outputtables.htmhttp://www.oecd.org/std/its/

  • Christian Dustmann, Bernd Fitzenberger, Uta Schnberg, and Alexandra Spitz-Oener 177

    The principle of autonomy of wage bargaining is laid down in the German The principle of autonomy of wage bargaining is laid down in the German constitution and implies that negotiations take place without the government constitution and implies that negotiations take place without the government directly exerting infl uence. As such, Germany has had no statutory minimum wage directly exerting infl uence. As such, Germany has had no statutory minimum wage imposed by the political process over the period we study. Rather, an elaborate imposed by the political process over the period we study. Rather, an elaborate system of wage fl oors is negotiated periodically between trade unions and employer system of wage fl oors is negotiated periodically between trade unions and employer associations, typically at the industry and regional level.associations, typically at the industry and regional level.

    This model of industrial relations has been very successful in Germany, where This model of industrial relations has been very successful in Germany, where negotiation with unions and participation of work councils in decision-making negotiation with unions and participation of work councils in decision-making processes are widely regarded as an important cornerstone in furthering processes are widely regarded as an important cornerstone in furthering common interests and even improving productivity. As a consequence, negotia-common interests and even improving productivity. As a consequence, negotia-tions are usually far more consensus-based and less confrontational than in other tions are usually far more consensus-based and less confrontational than in other countries. For example, Germany lost on average 11days of work each year per countries. For example, Germany lost on average 11days of work each year per 1,000employees by strikes and lock-outs between 1991 and 1999, but only fi vedays 1,000employees by strikes and lock-outs between 1991 and 1999, but only fi vedays per 1,000employees between 2000 and 2007. These fi gures for theearlier and later per 1,000employees between 2000 and 2007. These fi gures for theearlier and later time period compare to 40 and 32days per 1,000employees in the United States, time period compare to 40 and 32days per 1,000employees in the United States, 30 and 30days in the United Kingdom, 73 and 103days in France, 158 and 93days 30 and 30days in the United Kingdom, 73 and 103days in France, 158 and 93days in Italy, and 220 and 164days in Canada (Lesch 2009).in Italy, and 220 and 164days in Canada (Lesch 2009).

    Germanys culture of common interest is dissimilar to the view about worker Germanys culture of common interest is dissimilar to the view about worker representations commonly held in the United States. A recent US example is the representations commonly held in the United States. A recent US example is the attempt of the management of the German company Volkswagen to introduce attempt of the management of the German company Volkswagen to introduce a works council at its Chattanooga plant in Tennessee. While the participation of a works council at its Chattanooga plant in Tennessee. While the participation of works councils in management decisions is considered by Volkswagen as a corner-works councils in management decisions is considered by Volkswagen as a corner-stone of successful fi rm policy that helps furthering common interests, Tennessee stone of successful fi rm policy that helps furthering common interests, Tennessee Governor Bill Haslam has been outspoken in opposing any union formation at the Governor Bill Haslam has been outspoken in opposing any union formation at the plant, fearing that it endangers the states effort to attract investment (Greenhouse plant, fearing that it endangers the states effort to attract investment (Greenhouse 2013). A key difference between US and German labor market institutions lies in the 2013). A key difference between US and German labor market institutions lies in the fact that a works council in Germany elected by the employees does not have to be a fact that a works council in Germany elected by the employees does not have to be a union representative (although in practice the majority of works councils are union union representative (although in practice the majority of works councils are union representatives), while the installation of a works council in a US fi rm automatically representatives), while the installation of a works council in a US fi rm automatically involves the fi rm becoming unionized. Thus, works councils in Germany may act in involves the fi rm becoming unionized. Thus, works councils in Germany may act in greater independence from a union if the survival of their fi rm is at stake.greater independence from a union if the survival of their fi rm is at stake.

    Unions and Employer AssociationsIn Germany, contractual agreements between unions and employer associations In Germany, contractual agreements between unions and employer associations

    are negotiated either on the region-industry level or on the fi rm level. In addition to are negotiated either on the region-industry level or on the fi rm level. In addition to wages, working time regulations are an important component of the negotiations.wages, working time regulations are an important component of the negotiations.

    A distinguishing feature from US labor market institutions is that the recogni-A distinguishing feature from US labor market institutions is that the recogni-tion of trade unions in Germany is at the discretion of the fi rm, and union contracts tion of trade unions in Germany is at the discretion of the fi rm, and union contracts cover only the workers in fi rms that recognize the relevant sectoral wage bargaining cover only the workers in fi rms that recognize the relevant sectoral wage bargaining (union) contractregardless of whether the worker is a union member (for discus-(union) contractregardless of whether the worker is a union member (for discus-sion, see OECD 2004; Dustmann and Schoenberg 2009; Fitzenberger, Kohn, and sion, see OECD 2004; Dustmann and Schoenberg 2009; Fitzenberger, Kohn, and Lembcke 2013). Also, German fi rms that once recognized the union contracts can Lembcke 2013). Also, German fi rms that once recognized the union contracts can later opt out at their own discretion. Even within union wage contracts negotiated later opt out at their own discretion. Even within union wage contracts negotiated at the industry level, there is scope for wage fl exibility at the fi rm level through at the industry level, there is scope for wage fl exibility at the fi rm level through

  • 178 Journal of Economic Perspectives

    so-called opening or hardship clauses, provided that workers representatives so-called opening or hardship clauses, provided that workers representatives agree (for example, Hassel 1999; Hassel and Rehder 2001; Carlin and Soskice 2009; agree (for example, Hassel 1999; Hassel and Rehder 2001; Carlin and Soskice 2009; Brndle, Heinbach, and Meier 2011; Bispinck, Dribbusch, and Schulten 2010). Brndle, Heinbach, and Meier 2011; Bispinck, Dribbusch, and Schulten 2010). After opting out of a collective agreement, fi rms still have to pay wages for the After opting out of a collective agreement, fi rms still have to pay wages for the incumbent employees according to the collective agreement until a new agreement incumbent employees according to the collective agreement until a new agreement at the fi rm level has been reached, but they do not have to honor new negotiated at the fi rm level has been reached, but they do not have to honor new negotiated wage increases and the fi rm need not follow the old collective agreements for new wage increases and the fi rm need not follow the old collective agreements for new hires. Thus, over time a fi rm may be able to lower wage costs considerably by opting hires. Thus, over time a fi rm may be able to lower wage costs considerably by opting out of the union contractprovided its employees accepted this.out of the union contractprovided its employees accepted this.

    After 1995, there was indeed a dramatic decline in union coverage in Germany. After 1995, there was indeed a dramatic decline in union coverage in Germany. This decline is almost entirely driven by a decline in industry-wide agreements.This decline is almost entirely driven by a decline in industry-wide agreements.88 From 1995 to 2008, the share of employees covered by industry-wide agreements From 1995 to 2008, the share of employees covered by industry-wide agreements fell from 75 to 56percent, while the share covered by fi rm-level agreements fell fell from 75 to 56percent, while the share covered by fi rm-level agreements fell from 10.5 to 9percent. The percentage of German workers that were not covered from 10.5 to 9percent. The percentage of German workers that were not covered by an agreement in 1995 1997 was highest in the tradable services (22 percent), as by an agreement in 1995 1997 was highest in the tradable services (22 percent), as compared to tradable manufacturing (9.8percent) and nontradables (12percent). compared to tradable manufacturing (9.8percent) and nontradables (12percent). By 2006 2007, noncoverage had sharply increased in all threesectors to 40, 27, and By 2006 2007, noncoverage had sharply increased in all threesectors to 40, 27, and 32percent in the tradable services, manufacturing, and nontradables respectively, 32percent in the tradable services, manufacturing, and nontradables respectively, and this share continued to rise. By 2010, according to the German Structure of and this share continued to rise. By 2010, according to the German Structure of Earnings Survey, 41percent of all employees in fi rms with at least 10employees in Earnings Survey, 41percent of all employees in fi rms with at least 10employees in the sectors Manufacturing, Mining, and Services are not covered by any collective the sectors Manufacturing, Mining, and Services are not covered by any collective wage agreement (StaBu 2013).wage agreement (StaBu 2013).

    Has this decrease in union coverage rates contributed to a reduction in wage Has this decrease in union coverage rates contributed to a reduction in wage growth and to an increase in inequality? We investigate this question in Figure5, growth and to an increase in inequality? We investigate this question in Figure5, where we plot the observed changes in log real wages between 1995 and 2008 along where we plot the observed changes in log real wages between 1995 and 2008 along the wage distribution. We also plot the counterfactual changes that would have the wage distribution. We also plot the counterfactual changes that would have occurred if unionization rates had remained at the same level as in 1995, using the occurred if unionization rates had remained at the same level as in 1995, using the reweighting approach developed in DiNardo, Fortin, and Lemieux (1996), which reweighting approach developed in DiNardo, Fortin, and Lemieux (1996), which essentially reweighs wages observed in 2008 with the odds-ratio that a worker with essentially reweighs wages observed in 2008 with the odds-ratio that a worker with specifi c observed characteristics has been observed in the 2008-coverage-status in specifi c observed characteristics has been observed in the 2008-coverage-status in 1995 versus being observed in the 2008-coverage-status in 2008. Notice that this 1995 versus being observed in the 2008-coverage-status in 2008. Notice that this constructed counterfactual exercise is by no means causal, among other reasons constructed counterfactual exercise is by no means causal, among other reasons because it ignores general equilibrium effects of de-unionization. The fi gure because it ignores general equilibrium effects of de-unionization. The fi gure suggests that Germanys wages in 2008 would have been higher if union coverage suggests that Germanys wages in 2008 would have been higher if union coverage had remained the same as in 1995 throughout the entire wage distribution, but the had remained the same as in 1995 throughout the entire wage distribution, but the difference is particularly large at the lower end of the wage distribution.difference is particularly large at the lower end of the wage distribution.

    Works Councils and Opening ClausesWage inequality has also increased strongly among employees covered by union Wage inequality has also increased strongly among employees covered by union

    contracts, thus suggesting that the German system of industrial relations has allowed contracts, thus suggesting that the German system of industrial relations has allowed for wage adjustments even within the unionized sector. This pattern is illustrated in for wage adjustments even within the unionized sector. This pattern is illustrated in

    8 See Data AppendixA and TableA3 in AppendixC, available online with the paper at http://e-jep.org.

    http://e-jep.org

  • From Sick Man of Europe to Economic Superstar: Germanys Resurgent Economy 179

    Figure6, where we show the evolution of the 15th, 50th, and 85thpercentile of the Figure6, where we show the evolution of the 15th, 50th, and 85thpercentile of the wage distributions, indexed at 0 in 1995, for those covered by a union (panelA) and wage distributions, indexed at 0 in 1995, for those covered by a union (panelA) and uncovered by a union (panelB) sectors. The fi gure shows that wage inequality rose uncovered by a union (panelB) sectors. The fi gure shows that wage inequality rose strongly in the covered sector both at the bottom and the top of the wage distribu-strongly in the covered sector both at the bottom and the top of the wage distribu-tion, while in the uncovered sector it remained basically constant at the bottom tion, while in the uncovered sector it remained basically constant at the bottom of the wage distribution and only increased at the top of the wage distribution. of the wage distribution and only increased at the top of the wage distribution. However, notice that due to the indexation the fi gures hide the larger differentials However, notice that due to the indexation the fi gures hide the larger differentials in wage levels at specifi c percentiles in the uncovered sector relative to the covered in wage levels at specifi c percentiles in the uncovered sector relative to the covered sector: While the 8550th and 5015thdifferentials were on average 0.4 and 0.34 sector: While the 8550th and 5015thdifferentials were on average 0.4 and 0.34 in the covered sector between 1995 and 2008, they were about 0.5 in the uncovered in the covered sector between 1995 and 2008, they were about 0.5 in the uncovered sector. Thus, threefactors contributed to the rise in overall inequality during the sector. Thus, threefactors contributed to the rise in overall inequality during the time period under consideration, namely, the shift of workers from the covered to time period under consideration, namely, the shift of workers from the covered to the uncovered sector (which led, due to the larger differences in wage levels in the the uncovered sector (which led, due to the larger differences in wage levels in the uncovered sector, to an increase in lower tail inequality), the increase in inequality uncovered sector, to an increase in lower tail inequality), the increase in inequality in the covered sector, and the increase in inequality at the top of the wage distribu-in the covered sector, and the increase in inequality at the top of the wage distribu-tion in the uncovered sector.tion in the uncovered sector.

    Notes: The fi gure shows the observed wage growth by percentile between 1995 and 2008, as well as the counterfactual wage growth which would have prevailed if the share of workers covered either by industry-wide or fi rm-wide agreements had remained at its 1995 level. The counterfactual wage distribution is computed using the reweighting approach developed by DiNardo, Fortin, and Lemieux (1996). Calculations are based in the LIAB.

    Figure 5 Observed versus Counterfactual Wage Growth between 1995 and 2008 along the Wage Distribution: The Role of De-unionization

    Ch

    ange

    in lo

    g re

    al w

    age

    .1

    0

    .1

    .2

    5 15 25 857565554535

    Percentile

    Observed change

    Unionization at 1995 level

  • 180 Journal of Economic Perspectives

    Notes: Calculations based on LIAB Sample for West German Full-Time Workers between 20 and 60years of age. The fi gure shows the indexed (log) real wage growth of the 15th, 50th, and 85thpercentiles of the wage distribution, with 1995 as the base year. Nominal wages are defl ated using the consumer price index (1995 = 100) provided by the German Federal Statistical Offi ce.

    Figure 6Indexed Wage Growth of the 15th, 50th, 85th Percentiles, West Germany, by Union Coverage Status, 19952008

    Wag

    e gr

    owth

    .15

    .1

    .05

    0

    .05

    .1

    Wag

    e gr

    owth

    .15

    .1

    .05

    .05

    .1

    1995 1998 2001 2004 2007

    15th percentile

    50th percentile

    85th percentile

    A: Covered Sector

    85th percentile

    50th percentile 15th percentile

    B: Uncovered Sector

    Year

    1995 1998 2001 2004 2007

    Year

  • Christian Dustmann, Bernd Fitzenberger, Uta Schnberg, and Alexandra Spitz-Oener 181

    We think that the change in wage inequality in the covered sector is due We think that the change in wage inequality in the covered sector is due to the decentralization of wage setting since the beginning of the 1990s, when to the decentralization of wage setting since the beginning of the 1990s, when industry-level collective bargaining came under increasing pressure from employers industry-level collective bargaining came under increasing pressure from employers who demanded more fi rm-specifi c and differentiated regulations. Works councils who demanded more fi rm-specifi c and differentiated regulations. Works councils accommodated this decentralization to secure jobs in Germany, which also strength-accommodated this decentralization to secure jobs in Germany, which also strength-ened their role in the industrial relations. As another response, the trade unions ened their role in the industrial relations. As another response, the trade unions and employers associations agreed on an increasing number of opening clauses and employers associations agreed on an increasing number of opening clauses in industry-level collective agreements. Opening clauses allow fi rms to deviate from in industry-level collective agreements. Opening clauses allow fi rms to deviate from collectively agreed industry-wide standards. At fi rst, these opening clauses focused collectively agreed industry-wide standards. At fi rst, these opening clauses focused on hours of work, but later they also affected wages. Also, the opening clauses were on hours of work, but later they also affected wages. Also, the opening clauses were initially only temporary to avoid bankruptcy, but later they were also implemented initially only temporary to avoid bankruptcy, but later they were also implemented to ensure competitiveness in more general terms. Firms that use opening clauses to ensure competitiveness in more general terms. Firms that use opening clauses negotiate the details concerning pay and working time agreements with the works negotiate the details concerning pay and working time agreements with the works council. Under German law, fi rms without a works council cannot use opening council. Under German law, fi rms without a works council cannot use opening clauses, but such fi rms may instead decide to stop recognizing a union contract. clauses, but such fi rms may instead decide to stop recognizing a union contract. Firms with a works council not covered by a union contract may reach an agreement Firms with a works council not covered by a union contract may reach an agreement on wages with the works council.on wages with the works council.

    Brndle, Heinbach, and Meier (2011, Figure1) report that opening clauses for Brndle, Heinbach, and Meier (2011, Figure1) report that opening clauses for wages only started to gain importance in 1995 (opening clauses regarding hours of wages only started to gain importance in 1995 (opening clauses regarding hours of work had existed before 1995). Among industry-wide collective contracts in manu-work had existed before 1995). Among industry-wide collective contracts in manu-facturing, less than 5percent involved opening clauses for wages in 1995, but this facturing, less than 5percent involved opening clauses for wages in 1995, but this had risen to about 60percent in 2004. According to a survey of works councils in had risen to about 60percent in 2004. According to a survey of works councils in 2005, about 75percent of all fi rms with collective agreements use opening clauses 2005, about 75percent of all fi rms with collective agreements use opening clauses (Bispinck 2007; Bispinck, Dribbusch, and Schulten 2010).(Bispinck 2007; Bispinck, Dribbusch, and Schulten 2010).

    To summarize, the specifi c governance structure of the German system of To summarize, the specifi c governance structure of the German system of industrial relations allowed for an unprecedented increase in the decentralization industrial relations allowed for an unprecedented increase in the decentralization of the wage setting process, leading to a decrease in real wages, in particular at the of the wage setting process, leading to a decrease in real wages, in particular at the lower end of the wage distribution. This was driven by two main developments: lower end of the wage distribution. This was driven by two main developments: 1) a sharp decline in the share of workers covered by union agreements; and 1) a sharp decline in the share of workers covered by union agreements; and 2)an increase in opening clauses that strengthened the role of fi rm-based works 2)an increase in opening clauses that strengthened the role of fi rm-based works councils in wage determination relative to trade unions. This argument is consis-councils in wage determination relative to trade unions. This argument is consis-tent with the fi nding that the rise in fi rm-level differences in wages contributes tent with the fi nding that the rise in fi rm-level differences in wages contributes strongly to the rise in wage inequality in Germany (Antonczyk, Fitzenberger, and strongly to the rise in wage inequality in Germany (Antonczyk, Fitzenberger, and Sommerfeld 2010; Card, Heining, and Kline 2013).Sommerfeld 2010; Card, Heining, and Kline 2013).

    What Led to Greater Flexibility in the German Labor Market?

    Why were wage restraints and decreasing real wages at the lower end of the Why were wage restraints and decreasing real wages at the lower end of the wage distribution in Germany possible after the mid 1990s but not before? After wage distribution in Germany possible after the mid 1990s but not before? After all, German fi rms have always had the option not to recognize a union contract all, German fi rms have always had the option not to recognize a union contract and to pay wages below the union wage, provided their employees accepted this. and to pay wages below the union wage, provided their employees accepted this. Opening clauses had been possible before the mid 1990s. Our answer traces Opening clauses had been possible before the mid 1990s. Our answer traces to the major changes in Germanys economy in the early 1990s related to the to the major changes in Germanys economy in the early 1990s related to the

  • 182 Journal of Economic Perspectives

    reunifi cation of Germany and the opening up of the nearby central and eastern reunifi cation of Germany and the opening up of the nearby central and eastern Europeaneconomies.Europeaneconomies.

    On one hand, the extraordinary cost of German unifi cation burdened the On one hand, the extraordinary cost of German unifi cation burdened the German economy in an unprecedented way, which is partly responsible for Germanys German economy in an unprecedented way, which is partly responsible for Germanys dismal performance throughout the 1990s and early 2000s. The German Council of dismal performance throughout the 1990s and early 2000s. The German Council of Economic Experts (SVR 2004, table 100, p. 644) estimates net transfers of about Economic Experts (SVR 2004, table 100, p. 644) estimates net transfers of about 900billion euros from former West Germany to East Germany during the time period 900billion euros from former West Germany to East Germany during the time period 1991 to 2003. The total sum of net transfers corresponds to about half of oneyears 1991 to 2003. The total sum of net transfers corresponds to about half of oneyears GDP in Germany during that time period. On the other hand, the opening of central GDP in Germany during that time period. On the other hand, the opening of central and eastern European countries constituted a unique opportunity for German and eastern European countries constituted a unique opportunity for German industry to move production abroad. They offered a stable investment climate, as industry to move production abroad. They offered a stable investment climate, as well as (despite being locked away for several decades behind the Iron Curtain) a well as (despite being locked away for several decades behind the Iron Curtain) a long history of trade and interaction with Germany. The structure of industry and long history of trade and interaction with Germany. The structure of industry and education systems, for instance, shared many similarities, which survived the Soviet education systems, for instance, shared many similarities, which survived the Soviet era. Vocational training plays a key role in the education system, in a way similar to era. Vocational training plays a key role in the education system, in a way similar to Germany, in countries like Hungary or Poland. German was also widely spoken in Germany, in countries like Hungary or Poland. German was also widely spoken in parts of Central and Eastern Europe. At the same time, wages in these countries were parts of Central and Eastern Europe. At the same time, wages in these countries were far lower than in Germany, and working regulations more fl exible (for example, far lower than in Germany, and working regulations more fl exible (for example, Geishecker 2006; Marin 2006). Moving production abroad to these countries took Geishecker 2006; Marin 2006). Moving production abroad to these countries took place at a moderate pace: for example, the stock of German foreign direct invest-place at a moderate pace: for example, the stock of German foreign direct invest-ment to Poland, Hungary, as well as the Czech and the Slovak Republics amounted to ment to Poland, Hungary, as well as the Czech and the Slovak Republics amounted to about 1percent of German GDP in 2000 and about 2.3percent in 2010 (according to about 1percent of German GDP in 2000 and about 2.3percent in 2010 (according to our calculations and data from http://stats.oecd.org/Index.aspx?DatasetCode=FDIour calculations and data from http://stats.oecd.org/Index.aspx?DatasetCode=FDI_POSITION_PARTNER). However, the possibility that German fi rms might relocate _POSITION_PARTNER). However, the possibility that German fi rms might relocate production to these low-wage countries was very credible, and widely discussed in production to these low-wage countries was very credible, and widely discussed in public (among German media outlets, see the articles in public (among German media outlets, see the articles in DIHK 2003; Mihm and 2003; Mihm and Knop 2004; Hawranek, Hornig, and Jung2004).Knop 2004; Hawranek, Hornig, and Jung2004).

    The fi scal burden of German reunifi cation, coupled with an immediately The fi scal burden of German reunifi cation, coupled with an immediately more competitive global environment, made it increasingly costly for German more competitive global environment, made it increasingly costly for German fi rms to pay high union wages. The new opportunities to move production abroad, fi rms to pay high union wages. The new opportunities to move production abroad, while remaining still nearby, changed the power equilibrium between trade unions while remaining still nearby, changed the power equilibrium between trade unions and employer federations, and forced unions and/or works councils to accept and employer federations, and forced unions and/or works councils to accept deviations from industry-wide agreements which often resulted in lower wages deviations from industry-wide agreements which often resulted in lower wages for workers. In a similar vein, Burda (2000) predicted that the EU-accession of for workers. In a similar vein, Burda (2000) predicted that the EU-accession of Eastern European countries would foster a reduction of labor market rigidities in Eastern European countries would foster a reduction of labor market rigidities in the old EUmember countries (including Germany). Germanys unions and works the old EUmember countries (including Germany). Germanys unions and works councils realized that they had to make concessions in order not to be further councils realized that they had to make concessions in order not to be further marginalized, and the specifi c characteristics of the German system of industrial marginalized, and the specifi c characteristics of the German system of industrial institutions allowed the trade unions to adapt to the new economic realities and to institutions allowed the trade unions to adapt to the new economic realities and to make these concessions. As a result, the German labor market appeared to be far make these concessions. As a result, the German labor market appeared to be far more fl exible than many would ever have expected.more fl exible than many would ever have expected.

    Why did other continental European countries not react in the same way as Why did other continental European countries not react in the same way as Germany? One important reason is that the particularly diffi cult economic situa-Germany? One important reason is that the particularly diffi cult economic situa-tion in which Germany found itself in the early 1990s was to a large part specifi c to tion in which Germany found itself in the early 1990s was to a large part specifi c to

    http://stats.oecd.org/Index.aspx?DatasetCode=FDI_POSITION_PARTNER

  • From Sick Man of Europe to Economic Superstar: Germanys Resurgent Economy 183

    Germany, due to the reunifi cation of Germany, which was not felt in other European Germany, due to the reunifi cation of Germany, which was not felt in other European countries. This was reinforced by Germanys geographic vicinity to the countries of countries. This was reinforced by Germanys geographic vicinity to the countries of central and eastern Europe, which gave Germany an early taste of the challenges central and eastern Europe, which gave Germany an early taste of the challenges of globalization. This decade of economic stagnation and hardship, when Germany of globalization. This decade of economic stagnation and hardship, when Germany was the sick man of Europe, prepared the population for accepting agreements was the sick man of Europe, prepared the population for accepting agreements for the sake of economic growth, which saw inequality rise dramatically for the fi rst for the sake of economic growth, which saw inequality rise dramatically for the fi rst time in the after-war period.time in the after-war period.

    In addition, the system of industrial relations in other continental European In addition, the system of industrial relations in other continental European countries does not allow for the same inherent opportunities of fl exible adaptation countries does not allow for the same inherent opportunities of fl exible adaptation as the German system. For example, in countries like France and Italy, union wages as the German system. For example, in countries like France and Italy, union wages are often bargained at the national level and apply to all fi rms in the economy, are often bargained at the national level and apply to all fi rms in the economy, regardless of whether the fi rm explicitly recognizes the union contract. Coverage regardless of whether the fi rm explicitly recognizes the union contract. Coverage by union wage contracts has remained remarkably stable at very high levels at about by union wage contracts has remained remarkably stable at very high levels at about 90percent in France and 80percent in Italy during the 1990s and the 2000s (OECD 90percent in France and 80percent in Italy during the 1990s and the 2000s (OECD 2004, 2012; Visser 2013). Furthermore, in contrast to Germany, union wage contracts 2004, 2012; Visser 2013). Furthermore, in contrast to Germany, union wage contracts are typically extended to all workers in an industry (OECD 2004, table3.4, p.148; are typically extended to all workers in an industry (OECD 2004, table3.4, p.148; Visser 2013, table4, pp.96 98). In these and other continental European countries, Visser 2013, table4, pp.96 98). In these and other continental European countries, adding fl exibility into collective agreements would require political reforms at the adding fl exibility into collective agreements would require political reforms at the national level. More generally, many of the regulations which are determined by national level. More generally, many of the regulations which are determined by labor contracts in Germany are either legally enforced in other countries (such labor contracts in Germany are either legally enforced in other countries (such as the minimum wage in France) or nationally implemented (for example, union as the minimum wage in France) or nationally implemented (for example, union agreements extend to all fi rms in the economy), and therefore require consent agreements extend to all fi rms in the economy), and therefore require consent on a much higher level (nationally, or even on the political level) to be modifi ed on a much higher level (nationally, or even on the political level) to be modifi ed and changed. There is much less scope in these countries for a decentralization and changed. There is much less scope in these countries for a decentralization of wage setting (and other aspects of working conditions) within their system of of wage setting (and other aspects of working conditions) within their system of industrialrelations.industrialrelations.

    In general, the decentralization of union agreements is certainly being In general, the decentralization of union agreements is certainly being discussed more widely across Europe, but whether or when such changes might discussed more widely across Europe, but whether or when such changes might occur more widely remains uncertain.occur more widely remains uncertain.

    Discussion and OutlookDiscussion and Outlook

    We have argued that the remarkable transformation of the German economy We have argued that the remarkable transformation of the German economy from the sick man of Europe to a lean and highly competitive economy within from the sick man of Europe to a lean and highly competitive economy within little more than a decade is rooted in the inherent fl exibility of the German system little more than a decade is rooted in the inherent fl exibility of the German system of industrial relations. This system allowed German industry to react appropriately of industrial relations. This system allowed German industry to react appropriately and fl exibly over time to the demands of German unifi cation, and the global chal-and fl exibly over time to the demands of German unifi cation, and the global chal-lenges of a new world economy. However, this intrinsic fl exibility became only lenges of a new world economy. However, this intrinsic fl exibility became only evident under the extraordinary diffi cult economic circumstances and the extreme evident under the extraordinary diffi cult economic circumstances and the extreme duress in which Germany found itself in the decade after reunifi cation. How does duress in which Germany found itself in the decade after reunifi cation. How does our thesis fi t with two other possible explanations for Germanys increased competi-our thesis fi t with two other possible explanations for Germanys increased competi-tiveness: Germanys Hartz labor market reforms of 2003, or the changes brought tiveness: Germanys Hartz labor market reforms of 2003, or the changes brought about by the adoption of the euro?about by the adoption of the euro?

  • 184 Journal of Economic Perspectives

    Germanys government under Gerhard Schrder implemented the so-called Germanys government under Gerhard Schrder implemented the so-called Hartz Reforms to the labor markets in 2003, which are often credited for spur-Hartz Reforms to the labor markets in 2003, which are often credited for spur-ring Germanys economy (for example, Rinne and Zimmermann 2012, 2013; see ring Germanys economy (for example, Rinne and Zimmermann 2012, 2013; see Fitzenberger 2009 for a critical assessment of the Hartz Reforms). These reforms Fitzenberger 2009 for a critical assessment of the Hartz Reforms). These reforms were extremely controversial at the time. They reduced and limited the benefi ts while were extremely controversial at the time. They reduced and limited the benefi ts while unemployed, liberalized agency work, reformed active labor market policies, and unemployed, liberalized agency work, reformed active labor market policies, and reorganized the Federal Labor Agency, but did not make any institutional changes in reorganized the Federal Labor Agency, but did not make any institutional changes in the wage settingprocess.the wage settingprocess.

    The Hartz reforms were implemented starting in 2003, hence nearly a decade The Hartz reforms were implemented starting in 2003, hence nearly a decade after the process of wage decentralization and the improvement in competitive-after the process of wage decentralization and the improvement in competitive-ness had begun in Germany. It seems plausible that the changes already underway ness had begun in Germany. It seems plausible that the changes already underway in Germanys labor markets helped in preparing the political ground for the Hartz in Germanys labor markets helped in preparing the political ground for the Hartz reforms. In addition, as the enumeration of the main components of the reforms reforms. In addition, as the enumeration of the main components of the reforms makes clear, the scale of the reforms is modest enough that they seem unlikely to makes clear, the scale of the reforms is modest enough that they seem unlikely to have triggered the dramatic increase in competitiveness or the enormous drop in have triggered the dramatic increase in competitiveness or the enormous drop in German unemployment or to have led Germanys labor market through the deep German unemployment or to have led Germanys labor market through the deep recession in 20082009. Further, while the focus of the reforms was on creating recession in 20082009. Further, while the focus of the reforms was on creating incentives for seeking employment, they did little to support the remarkable wage incentives for seeking employment, they did little to support the remarkable wage restraint witnessed since the mid 1990s, which is the key factor in explaining the restraint witnessed since the mid 1990s, which is the key factor in explaining the gain in competitiveness.gain in competitiveness.

    We therefore believe that while the Hartz reforms have contributed to the We therefore believe that while the Hartz reforms have contributed to the recent decline in long-term unemployment and to the continued increase in wage recent decline in long-term unemployment and to the continued increase in wage inequality at the lower end of the wage distribution, they were not central or inequality at the lower end of the wage distribution, they were not central or essential in the process of improving the competitiveness of German industry. essential in the process of improving the competitiveness of German industry. Moreover, although one sometimes hears the argument that other continental Moreover, although one sometimes hears the argument that other continental European countries should muster the political will to adopt their own version European countries should muster the political will to adopt their own version of the Hartz reforms, we believe that such a recommendation may be misleading. of the Hartz reforms, we believe that such a recommendation may be misleading. In our view, the specifi c governance structure of the German system of indus-In our view, the specifi c governance structure of the German system of indus-trial relationsactivated under extreme duressis what paved the way for the trial relationsactivated under extreme duressis what paved the way for the remarkable decentralization of wage determination from the industry level to remarkable decentralization of wage determination from the industry level to thelevel of the single fi rm or single worker, and which together with a signifi cant thelevel of the single fi rm or single worker, and which together with a signifi cant increase in productivity ultimately improved Germanys competitiveness. Whether increase in productivity ultimately improved Germanys competitiveness. Whether the political process would have been able to achieve a similar degree of wage the political process would have been able to achieve a similar degree of wage decentralization, had the autonomy of wage bargaining not existed in Germany, decentralization, had the autonomy of wage bargaining not existed in Germany, is doubtful. In our view, the policy recommendation from Germany for the rest is doubtful. In our view, the policy recommendation from Germany for the rest of continental Europe should not be the Hartz reforms (the advice given often by of continental Europe should not be the Hartz reforms (the advice given often by policymakers, as in a February 2013 speech by German Chancellor Angela Merkel policymakers, as in a February 2013 speech by German Chancellor Angela Merkel reported in de Weck 2013), but reforms that would target the system of industrial reported in de Weck 2013), but reforms that would target the system of industrial relations by decentralizing bargaining to the fi rm level while keeping workers relations by decentralizing bargaining to the fi rm level while keeping workers representatives involved to secure that employees benefi t again when economic representatives involved to secure that employees benefi t again when economic conditions improve.conditions improve.

    Some argue that the adoption of the common European currency is a main Some argue that the adoption of the common European currency is a main factor that has helped Germany to improve competitiveness. Again, we believe that factor that has helped Germany to improve competitiveness. Again, we believe that the arrival of the euro may have been a contributing factor, but not the main one. the arrival of the euro may have been a contributing factor, but not the main one.

  • Christian Dustmann, Bernd Fitzenberger, Uta Schnberg, and Alexandra Spitz-Oener 185

    First, recall that Germany was shifting its labor market institutions and improving First, recall that Germany was shifting its labor market institutions and improving its compet